Overview

Role: Director of Product Design
‍Team: Product Design Team
‍Timeline: 2024-2025

Purpose

Purpose of the AI Assistant is to provide impactful support to user workflows as a complement to existing AI Console features across the Interact platform (platform used by IPG for end-to-end media lifeycle management). There were no pre-existing workflows and we needed to research, produce a proof of concept and scope designs for an AI Assistant MVP that would scale well across all products in the platform.

Team

Selected and led designers working on this effort

• Michal Polak, Associate Product Design Director
• Mariana Savitska, Senior Product Designer
• Dominik Sfluvik, Product Designer
• Oleksandr Ladieiev, Junior Product Designer

Process

Discovery & Ideation

• Led a series of whiteboarding sessions with product and design teams to align on AI integration opportunities
• Mapped out potential AI touchpoints across the platform to identify where the assistant could deliver the most value

Prototyping & Validation

• Designed early wireframes and prototypes for a "copilot experience" that introduced AI-assisted workflows
• Created a UI kit and flow diagrams to demonstrate scalability across different product areas
• Collected structured feedback from product teams to refine usability and technical feasibility
• Worked actively with designers to refine designs to get to an MVP

Prioritization & Complexity Assessment

• Provided insights to structure our feature roadmap, categorizing features by complexity
• Worked closely with engineering to evaluate implementation feasibility

Refinement & MVP Definition

• Finalized the MVP with designers, focusing on a floater window and contextual toolbar actions for seamless AI assistance
• Worked with designers to create a flexible and clear way to validate MVP with product, developing a scalable research template for highlighting key AI-driven workflows across all products

Discovery & Ideation

There had already been many discussions and ideas circulating across products organically about how to integrate AI Console capabilities into specific product workflows. To consolidate all the thinking relevant to AI Assistant, I led a series of discussions internally with the design team and with product collaborators to begin developing a point of view on how to integrate AI Assistant across all products.

Proof of Concept

Since this was a project with no pre-existing workflows or clearly defined scope at the beginning (up to us to clarify and define), I take a hands-on approach in designing wires and initial proof of concepts for the designers to align and polish for the final MVP.

In parallel with discovery and ideation, I drafted up wires for what what was initially circulating as a "copilot experience".

I also developed and designed the example UI kit, flow and prototypes for the proof of concept to circulate, share and gain more structured feedback across products on the experience and scalability

Product Sketches

Once the initial proof of concept began to take shape, I worked actively with designers to refine the design and created a template for how it would hypothetically be scaled across all products. We wanted to test the reaction of this experience across all product groups in the platform, so we used this as a template to gather input on usability, scalability, and ideas for how the AI Assistant would support product workflows.

Prioritization

By MVP Scope

As we began to see common priorities and key areas of impact take shape, we began to first focus the AI Assistant MVP on the "floater window", and contextual toolbar actions such as running smart actions on interaction with the interface second.

By Complexity

To share and begin evaluating build around the key priorities we were seeing emerge across all collaborators from the initial product sketches with engineering, I put together a summary of key ideas by complexity.

Lower Complexity

Information Retrieval & Basic Insights

Tasks focused on quickly retrieving and displaying relevant information using pre-defined rules and structured search functions. They require minimal AI customization and are easy to implement using existing data sources. Examples include:

Competitive Analysis

• Fetching competitor spending and activity insights
• Identifying strongest competitors in a given industry
• Displaying performance benchmarks based on industry data

Reporting & Insights

• Automating campaign performance summaries and reports
• Providing structured insights on media investments
• Generating quick responses to common analytics questions

Audience Insights

• Identifying top-performing audience segments
• Understanding customer pain points based on engagement data
• Highlighting emerging trends based on audience behaviors

Medium Complexity

Data Analysis & Recommendations

Medium complexity tasks involving data analysis, identifying patterns, and providing actionable recommendations. Examples include:

Campaign Strategy

• Assisting in campaign brief development
• Recommending key messaging elements based on past success
• Identifying target audience characteristics for optimized outreach

Budget Allocation

• Suggesting budget splits across different media channels
• Adjusting allocations based on performance and pacing trends
• Recommending shifts in investment to maximize ROI

Media Plan Optimization

• Recommending best-performing channels based on campaign goals
• Exploring different media mix scenarios to optimize impact
• Identifying inefficiencies in media spend and providing corrections

Higher Complexity

Generative AI & Advanced Automation

These tasks involve higher levels of AI-driven creativity, automation, and decision-making. They require significant customization, training, and integration with advanced/larger data sources. Examples include:

Creative Generation

• Generating images based on text prompts (e.g., "Create a 300x250 image of a car in the jungle")
• Producing brand-compliant creative templates for awareness campaigns
• Suggesting ad copy based on brand tone and campaign objectives

AI-driven Audience Creation

• Automatically generating audience segments based on user inputs
• Suggesting lookalike audiences similar to existing customers
• Identifying market gaps based on engagement patterns

Advanced Data Interpretation

• Providing human-like interpretations of complex data reports
• Enabling intelligent semantic search for emerging trends
• Offering deep investment insights and performance tracking

MVP Designs

For the initial MVP, we decided to focus on an extension view of the AI Assistant: a context-aware, on-demand assistant accessible via a persistent top navigation extension across the Interact platform. It enables users to interact with AI-powered insights and actions without disrupting their workflow.

Key Features

✅ On-Demand Accessibility

• Users can access the AI assistant at any time via the top navigation
• The floater window overlays the interface without persistent screen real estate usage

✅ Context-Aware Interactions

• Detects active workflows and provides relevant AI-powered assistance
• Suggests smart actions based on the user’s current task

✅ Quick AI Responses & Smart Actions

• Users can input queries or select from AI-suggested actions
• Supports information retrieval (e.g., campaign insights, competitor benchmarks)
• Automates common workflows, such as summarizing reports or generating media plans

✅ Adaptive UI & Seamless Integration

• The interface dynamically adjusts based on the user’s location within the platform
• Designed for non-intrusive, efficient interactions without disrupting ongoing tasks

✅ Scalable for Future Enhancements

• Built as a modular system, allowing for expansion with more AI-driven capabilities over time

Next Steps

✔ User Testing & Iteration: Conduct usability tests with our users to validate impact and direction

✔ Refinement of MVP Capabilities: Expand initial MVP capabilities to provide contextual recommendations and deeper analytical insights

✔ Technical Implementation: Partner with engineering to define a structured implementation plan and identify use cases for priorities

✔ Scalability Roadmap: Explore long-term AI integration strategies to support future product direction